xmlhack: Semantic Web to be Berners-Lee's second success?

A report from WWW10.

With the Semantic Web, Tim Berners-Lee is attempting to reproduce the technique that worked so well with the Web -- a simple concept magnified by billion of distributed sites. However, more effort will be needed to convince the skeptics.

Some WWW10 delegates, including W3C members from leading vendors, explained that they consider the Semantic Web a distraction for the W3C, and think that resources and focus should be employed to progress simple specifications enabling web services and e-commerce applications.

These protesters find allies amongst knowledge management specialists, who believe that the problem of knowledge representation addressed by the Semantic Web is too complex to be solved by simple solutions.

During his keynote, Tim Berners-Lee reminded the skeptics that the Web was built on a very simple adaptation of complex hyperlinking technologies, and its power comes from the number of participating sites. To convince them that the same process can be used to build the Semantic Web, he announced that he was resuming hacking, and presented some preliminary results during a W3C track.

Semantic Web applications that are not focused on the enterprise will take much longer to gain traction. Consumers cannot be relied upon to push a new technology, especially while the global economy is experiencing a slowdown.

Re: Semantic Web to be Berners-Lee's second success? (Eric van der Vlist - 06:39, 12 May 2001)

Re: Semantic Web to be Berners-Lee's second success? (Rappenecker - 16:52, 10 May 2001)

The four paragraphs above appear to be a very high-level summary of Tim's address. Where can the full text or content of his presentation be accessed?

Re: Semantic Web to be Berners-Lee's second success? (Kurt Cagle (Author: XML Developers Handbook) - 18:26, 7 May 2001)

Of course, it should be noted that many of the same semantic experts had been building complex languages like Prolog and SGML for years without hitting on the idea of making it ... um, usable. There is no doubt that the idea of a semantic web is more complex than a web of hyperlinks, and that such semantic information isn't necessarily as easy to build into the system as people would like. However, the operational term here is one of simplicity -- build a system that has the versamilitude of semantic awareness, even if it isn't fully aware, make it something that has a minimum number of parts -- not a five hundred page spec but a fifty page one, or better yet a five page one --, make it freely available, and it will grow in the direction it needs to.

I also take issue with those people who feel that the real concentration shouldn't be on the Semantic Web, but rather on building the infrastructure to make e-commerce possible. This is being pushed strongly by big moneyed interests, primarily software vendors, who are more interested in selling products (and locking in customers) than building community infrastructure. The irony is that many of the big standards that these companies are pushing: SOAP, UDDI, ebXML, etc., are crashing and burning all over the place; they are complex, slow, cumbersome, and require a level of cooperation between companies that is simply not present. As a consequence, you get SOAP messages which contain tiny payloads and huge authentication/encryption blocks, making for extremely processor intensive transactions that can already be handled far more efficiently with existing systems.

Meanwhile, you're beginning to see the evolution of the Semantic Web all on its own. RSS feeds, DocBook article systems, Jabber chats built on XML messages passing back and forth, the slow but steady acceptance of XSLT as the foundation of an architecture, the growing interest in RDF and topic maps, I can go on for a while. They won't make anybody rich, because they are technologies of the commons, but they will be around long after Microsoft and Sun and Oracle are dust. They ARE the semantic web, and no amount of Academic naysayers or corporate marketers will change that fact.

Kurt Cagle http://www.kurtcagle.net XSLT and Metaphorical Programming

Re: Semantic Web to be Berners-Lee's second success? (Phani Vaddadi - 05:31, 7 May 2001)

There is ample literature to support the assertion that it's not possible to represent semantic data in quantifiable way. Chpter12 in An introduction to Database systems by CJDate, Date brings covers the E/R model. E/R model is the equivalent of semantic modeling as it applies to the relational representation of data. That said there is a lot of value to the research that Tim Barners-Lee is spearheading. HTML and XML are indeed reinventing a lot of what the hierarchical and object oriented databases discovered. But everytime the reinvention happens it's applied in different ways. HTML is successful because of the fact that it let people design systems that are not fail safe. XML and the RDF technologies propose to do the same thing. The semantics so built while may not be understood by everyone, will still enable communities of agents with reasonably well understood semantics that can process data. This is a reasonable goal and is surprisingly achievable.

Re: Semantic Web to be Berners-Lee's second success? (Mike Champion - 20:01, 4 May 2001)

"These protesters find allies amongst knowledge management specialists, who believe that the problem of knowledge representation addressed by the Semantic Web is too complex to be solved by simple solutions. " I don't disagree, but can anyone provide a more specific reference or more detailed explanation?

Re: Semantic Web to be Berners-Lee's second success? (Len Bullard - 14:03, 4 May 2001)

If a great success means that every decade, we roll back and pillage the research of a prior decade, then implement it over new hardware, sure it will be a great success.

There aren't definitional problems to this initiative. Given all the experience of the previous decades, adapting that work for ontological metadata isn't very hard. Prolog taught us the hard bits. However experience also taught that the real problem of using automated reasoning and deduction lay in the sources of the information, their authority and legitimacy for claiming the definitions, and the uneven competence for building these, and the expense of their production and maintenance.

Ontological systems require production and testing constraints. This is not like simply building a web page with meta data. Analysis systems are pattern-seeking and pattern recognition requires a truth value before actions are taken based on the results of such analysis. Such systems have problems of superstitious assertion and gaming.

What the semantic web will replicate is a slightly more formal looking version of the web page in which both good assertions and bad assertions have been made public and are acted on. A simple universal system such as Berners Lee proposes solves the problem of access but not of authority and as trust is built over the system the potential to game that trust is built as well.

Re: Semantic Web to be Berners-Lee's second success? (bryan - 07:54, 4 May 2001)

although it sounds crazy perhaps specs should be limited to 100 pages, I mean the original xml was what? 48? much easier to read through and thus implement. they talk so much about modularizing everything. Start by modularizing the specifications.